The basolateral membrane sorting determinant of an inwardly rectifying potassium channel, Kir 2.3, is comprised of a unique arrangement of trafficking motifs containing tandem, conceivably overlapping, biosynthetic targeting and PDZ-based signals. In the present study, we elucidate a mechanism by which a PDZ interaction coordinates one step in a basolateral membrane sorting program. In contrast to apical missorting of channels lacking the entire sorting domain, deletion of the PDZ binding motif caused channels to accumulate into an endosomal compartment. Here, we identify a new human ortholog of aCaenorhabditis elegans PDZ protein, hLin-7b, that interacts with the COOH-terminal tail of Kir 2.3 in renal epithelia. hLin-7b associates with the channel as a part of a multimeric complex on the basolateral membrane similar to a basolateral membrane complex inC. elegans vulva progenitor cells. Coexpression of hLin-7b with Kir 2.3 dramatically increases channel activity by stabilizing plasma membrane expression. The discovery identifies one component of the sorting machinery and provides evidence for a retention mechanism in a hierarchical basolateral trafficking program.
- membrane trafficking
- intracellular sorting mechanism
- protein-protein interaction
- inward rectification
- cortical collecting duct
the polarized expression of disparate transport proteins on two distinct membrane domains is an essential prerequisite for the vectorial transport of water, solutes, and ions across epithelia. The renal cortical collecting duct (CCD), the site of potassium secretion in the kidney, provides a salient example. In these cells, the asymmetric expression of weakly inward-rectifying K+ channels on the apical membrane and more strongly rectifying K+ channels on the basolateral membrane may increase the fidelity of the secretory process, ensuring that potassium preferentially exits the cell across the apical membrane into the tubule lumen in concert with the demands of potassium homeostasis (10). The identification of a plausible gene candidate, Kir 2.3, for the basolateral K+channel (39) provided the impetus to elucidate the basis for polarized membrane targeting of a native channel (21). In this effort, we discovered a novel sorting motif that is responsible for coordinating basolateral membrane expression of the strong inwardly rectifying K+ channel, Kir 2.3 (22).
In renal epithelial cells, most newly synthesized membrane proteins are thought to be sorted at the level of the trans-Golgi network (TGN) to the appropriate cellular domains (5, 24). In these cases, directed trafficking is dictated by signals located within the membrane protein structure, usually located within the cytoplasmic COOH-terminal domains (5). The sorting machinery in the TGN reads, interprets, and acts on the signals to target molecules to their correct destinations. Once delivered to the appropriate domain, many of these integral membrane proteins are effectively anchored at these polarized locales through interactions with the cytoskeleton or other membrane-associated proteins, completing the polarization program (25, 41). Several different basolateral sorting signals have been identified. The recent discovery of a novel form of the AP-1 clathrin adaptor complex (AP-1b), containing an epithelial-specific μIb-subunit (27), provided a plausible basolateral sorting mechanism for at least one class of these signals (8). Indeed, the capacity of the μ-adaptins to interact with a basolateral sorting motif that is colinear with tyrosine-based, clathrin-dependent endocytotic signal (2) suggests that the AP-1b complex may be recruited to vesicles in the TGN by specific membrane protein cargo containing these interaction motifs, marking them for basolateral membrane trafficking.
The basolateral sorting motif in the Kir 2.3 channel appears to be distinct from the classic tyrosine-based, clathrin-dependent endocytotic signal (22), suggesting interaction with targeting machineries other than AP-1b. Interestingly, the sorting signal is located at the extreme COOH terminus and juxtaposes or is colinear with a PDZ binding motif. PDZ domains, named after the three homologous proteins in which they were first discovered [PSD95, Dlg (Disc large), and ZO-1 (zona occludens)], are 80- to 90-amino acid residue protein interaction modules that generally recognize specific short motifs found at the extreme COOH-terminal domain of certain proteins. These domains have been strongly implicated in bringing multiprotein complexes together to localize expression on particular membrane domains or intracellular compartments (12).
In the present study, we evaluated the role of the Kir 2.3 PDZ binding motif within the broader sorting domain. Here, we identify and characterize an ortholog of a Caenorhabditis elegans PDZ protein, hLin-7b, that binds to the COOH-terminal tail of Kir 2.3. The discovery identifies one component of the sorting machinery and provides compelling evidence for a retention mechanism in a hierarchical basolateral trafficking program.
MATERIALS AND METHODS
Yeast two-hybrid interaction trap library screen.
The yeast two-hybrid interaction trap system was used according to established methods (11, 13) to identify kidney proteins that interact with the basolateral sorting signal in the extreme COOH terminus of Kir 2.3. Specifically, a sequential, two-step library screen with two different baits was performed to specifically capture proteins that interact with the PDZ binding motif in Kir 2.3. In the first stage of the screen, a lexA fusion protein of the Kir 2.3 COOH-terminal targeting domain (amino acids 417–445) was used as the bait (pJK202 containing a HIS3-selectable marker) to screen a conditionally expressed kidney cDNA library (Clontech) in the pJG4–5 plasmid (TRP4-selectable marker). Saccharomyces cerevisiae (EGY48, Mat a-ura3, his3, trp1, ura3, 3lexAop-leu2) were transfected with the wild-type bait and the kidney cDNA library and then plated onto one of two different selection plates. The number of transfectants was determined by plating yeast in a drop-out medium that selects for the presence of each plasmid (uracil-, histidine-, tryptophan-). In this system, interaction of library and bait proteins causes the transcriptional activation of two different reporter genes, leu2 and lacZ. Subsequently, interacting clones were identified by plating yeast onto medium that contains X-gal and lacks leucine, uracil, histidine, and tryptophan (2% galactose and 1% raffinose as carbon sources). Yeast that survived and turned blue were considered positive. Plasmid DNA from positive yeast clones was collected, and library pJG4–5 plasmids were isolated. In the second stage of the screen, positive library clones were screened with a lexA fusion of a mutant Kir 2.3 COOH-terminal domain that lacks the PDZ binding motif (443X). Library clones that scored positively in the first round with the wild-type bait but failed to interact with the truncated bait were considered true positives. These clones were sequenced using pJG4–5 primers.
The vesicular stomatitis virus glycoprotein (VSV) epitope-tagged wild-type Kir 2.3 in the mammalian expression vector PCB6 was used as described previously (21). The PDZ ligand in Kir 2.3 was disrupted by introducing a premature stop codon at amino acid position 442 with PCR-based mutagenesis and cloned into the PCB6 vector. To specifically recognize hLin-7b in cells, we constructed an epitope-tagged version. The hemagglutinin protein (HA) epitope (YPYDVPDYA) was engineered into either the NH2 or COOH termini of hLin-7b by PCR and subcloned into the pcDNA3.1 vector (Invitrogen). Creation of wild-type and mutant Kir 2.3 COOH-terminal lexA fusion proteins was achieved by PCR-based mutagenesis and subsequent subcloning in frame with the lexA DNA binding domain in the pJK202 vector. Sequences were verified either by dye termination sequencing (ABI Prism) or with chain-terminating inhibitors using Sequenase.
Cell culture and transfections.
COS cells were grown in DMEM supplemented with 10% fetal bovine serum, 100 U/ml penicillin, 100 μg/ml streptomycin, and 10 mM HEPES. Madin-Darby canine kidney (MDCK) cells were grown in DMEM supplemented with 10% fetal bovine serum, 100 U/ml penicillin, 100 μg/ml streptomycin, and 2 mM l-glutamine. For transfections, cells at ∼70–80% confluence were washed once with serum-free DMEM and then incubated with 10 μl of Lipofectamine (GIBCO-BRL) and ∼1 μg of DNA in OptiMEM (GIBCO-BRL) for 5 h. After the initial incubation with the transfection mix, cells were supplemented at a 1:1 ratio with DMEM containing 20% fetal bovine serum and 4 mMl-glutamine. Transfection medium was replaced 24 h after transfection with maintenance medium supplemented with 2 mM sodium butyrate to enhance protein expression.
Stably transfected MDCK cell lines were established as previously described (21). All vectors used provide neomycin resistance. Seventy-two hours after transfection, MDCK cells were placed in DMEM containing 500 μg/ml G418. Surviving clonal cell colonies were isolated within 12–18 days of initial neomycin selection, expanded, and maintained in DMEM containing 500 μg/ml G418.
Immunoprecipitation and immunoblotting.
COS cells, at 70–85% confluence, were transfected with epitope-tagged Kir 2.3 and hLin-7b constructs with Lipofectamine according to the manufacturer's specifications. Confluent, stably transfected MDCK cells expressing HA-hLin-7b and transiently transfected COS cells were incubated in growth medium containing 2 mM sodium butyrate for 24 h before harvesting. Forty-eight hours after transfection, cells were washed once with ice-cold PBS, harvested in cold lysis buffer (150 mM NaCl, 20 mM Tris pH 7.5, and 5 mM EDTA), pelleted (2,000 g for 5 min), and resuspended (∼5× cell pellet volume) in lysis buffer containing 1% Triton X-100 and a protease inhibitor cocktail (10 μg/ml antipain, 10 μg/ml leupeptin, 1 mM phenylmethylsulfonyl fluoride, and 10 μg/ml pepstatin A). Cells were then passed though a 27-gauge needle, rotated at 4°C for 1 h, and then centrifuged at 15,000 g for 15 min at 4°C. Soluble fractions from COS cells were precleared using 100 μl of a 50% Sepharose B slurry at 4°C for 2 h with rotation. Precleared COS cell lysates were then rotated overnight with 100 μl of a 10% protein A slurry with or without 2 μg of rabbit anti-HA antibody (Santa Cruz Biotechnology). MDCK cells were incubated overnight with agarose-conjugated anti-HA antibody (Santa Cruz Biotechnology) at 4°C. After being washed three times with lysis buffer containing 0.1% Triton, the immunoprecipitates from COS and MDCK cell lysates were eluted for 30 min at room temperature with SDS sample buffer and 0.1 M glycine (pH 3.5), respectively. Eluates were separated on a 10% SDS-polyacrylamide gel and transferred to nitrocellulose (Amersham). For Western blotting, mouse anti-VSV (P5D4) (Sigma) and mouse anti-HA (Boehringer Mannheim) antibodies were used at 10 μg/ml and 1:1,000, respectively. Horseradish peroxidase (HRP)-conjugated goat anti-mouse antibody (Jackson Laboratories) was used at 1:10,000 for detection with the enhanced chemiluminescence system (Amersham). The mouse anti-CASK (Transduction Laboratories) and mouse anti-syntrophin (9) antibodies were both used at 1:500 and detected with HRP-conjugated goat anti-mouse antibody (Sigma) at a 1:2,000 dilution.
Immunofluorescence and confocal microscopy.
For immunolocalization studies, MDCK cells were plated onto Transwell filters (Costar) and grown until they formed a confluent monolayer. Confluent cells, incubated in medium supplemented with 2 mM sodium butyrate overnight, were washed twice with PBSCM (PBS + 1 mM CaCl2 + 1 mM MgCl2) and fixed for 30 min in 3.7% paraformaldehyde. Cells were then washed three times with PBSCM, quenched in PBSCM containing 50 mM NH4Cl for 30 min, and washed two times with PBSCM. MDCK cells were permeabilized with 0.1% Triton in PBSCM for 30 min, washed four times with PBSCM, and blocked with 1% BSA in PBSCM for 15 min. Cells were then incubated overnight at 4°C with the primary antibodies in PBSCM containing 0.1% BSA. The rabbit anti-HA (Santa Cruz Biotechnology) and mouse anti-VSV (P5D4) antibodies were used at 1:200 and 1:1,000, respectively. Mouse anti-CASK was used at 10 μg/ml. Cells were then washed three times with PBSCM and incubated with the secondary antibodies in 0.1% BSA-PBSCM for 30 min at room temperature. Alexa 488-conjugated goat anti-rabbit (Molecular Probes) and rhodamine-conjugated goat anti-mouse (Jackson Laboratories) antibodies were used at 1:100 and 1:80 dilutions, respectively. Labeled cells were rinsed four times with PBSCM and mounted onto slides with Vectashield glycerol (Vector Labs).
Colocalization of 442X channels with transferrin was performed using transferrin conjugated to biotin. Confluent MDCK cells were washed once with serum-free DMEM and then incubated for 1 h in serum-free DMEM containing 50 μg/ml biotin-conjugated transferrin (Sigma) at 37°C. Cells were then washed three times with PBS and fixed as described above. Fixed cells were washed three times with PBS and permeabilized with PBS containing 0.1% saponin for 30 min. Cells were then incubated with Texas red-conjugated streptavidin (Jackson Laboratories) at 3 μg/ml in PBS containing 0.1% saponin for 1 h at room temperature. Labeled cells were then washed three times with PBS and blocked with 1% BSA for 15 min. Immunolocalization of 442X channels was performed as above, except that the antibody incubations were performed in PBS containing 0.1% saponin.
Fixation of male 129/SvEv mouse kidneys was achieved by perfusion via the heart. To clear blood from the kidneys, PBS was perfused for 2 min before a 5- or 30-min perfusion with paraformaldehyde (2%) and a 2-min perfusion with a cryoprotectant (10% EDTA in 0.1 M Tris). Twelve-micrometer-thick kidney sections were cut using a cryostat, placed on coverslips, coated with HistoGrip (Zymed), and stored at −80°C. For immunolocalization, sections were first rehydrated with PBS and then treated with 6 M guanadine-HCl to unmask protein epitopes. Kidney sections were then washed three times in a high-salt buffer (PBS containing 1% BSA and 385 mM NaCl), blocked (PBS containing 1% BSA and 50 mM glycine), and incubated overnight with primary antibodies (10 μg/μl) at 4°C in PBS supplemented with 0.1% BSA and 0.02% NaN3. Room temperature washes with high-salt buffer were performed three times for 5 min, once for 15 min, and once for 30 min to remove nonspecific binding. Alexa 488- and 568-conjugated secondary antibodies (1:100) were incubated for 2 h at 4°C in PBS containing 0.1% BSA and 0.02% NaN3. Sections were then washed as described above, mounted onto slides in Vectashield, and sealed with nail polish.
To determine cellular localization of hLin-7b and Kir 2.3 constructs, cells were visualized using the 410 Zeiss laser-scanning microscope under a ×63 oil-immersion lens. Cell sections were taken at steps of 0.5 μm with a zoom of 2 and a pinhole size of 18. Color assignment to fluorescence was performed using Metamorph, and images were processed with Adobe Photoshop.
2-Nitrophenyl-β-d-galactopyranoside solution β-galactosidase assay and protein extraction from yeast.
To determine the relative strength of the interactions, β-galactosidase activity was measured at maximum velocity using the 2-nitrophenyl-β-d-galactopyranoside (ONPG) solution assay according to established methods (26). Because the interaction trap uses GAL1-inducible promoter to conditionally express the AD prey protein, yeast were grown (A600 of 0.9–1.2) under conditions that induce (2% galactose and 1% raffinose) or repress (2% glucose) the promoter. Yeast cultures were pelleted (2,000 g) and resuspended in 750 μl of Z buffer (in mM: 60 Na2HPO4, 40 NaH2PO4, 10 KCl, 1 MgSO4, and 50 β-mercaptoethanol). For the assay, 100 μl of the Z buffer cell suspension was added to 900 μl of Z buffer in glass culture tubes containing 10 μl of 0.1% SDS and 20 μl of chloroform. Tubes were vortexed for 15 s and equilibrated at 30°C for 15 min. After equilibration, 200 μl of stock 4 mg/ml ONPG was added to the samples and vortexed for 5 s. Reactions were stopped at 30 s to 4 min by adding 0.5 ml of 1 M Na2CO3 and centrifuged to pellet debris, and the reaction product was measured by spectrofluorometery. Miller units were calculated as previously described (30).
To ensure that all lexA-Kir 2.3C constructs were expressed at relatively equal levels, yeast were examined for expression of the different lexA fusion proteins. Cultures grown in the appropriate dropout medium (A600 ∼0.6) were spun down at 2,000g for 5 min in a Sorvall tabletop centrifuge at room temperature. Pellets were resuspended and lysed in 100 μl of cracking buffer (8 M urea, 5% SDS, 40 mM Tris · HCl pH 6.8, 0.1 mM EDTA, 0.4 mg/ml bromphenol blue, 125 mM β-mercaptoethanol, and protease inhibitor cocktail) per 7.5 OD600 units and 300 μl of glass beads, incubated at 70°C for 10 min, and vortexed for 1 min. Samples were centrifuged at 15,000 g at 4°C, and supernatants were collected and separated using SDS-PAGE. After the gel was transferred to nitrocellulose, the rabbit anti-lexA antibody (Clontech) was used at a concentration of 1 μg/ml for detection of the lexA fusion proteins.
Oocyte isolation and injection.
Oocytes were isolated from female Xenopus laevis (NASCO; Fort Atkinson, WI). After anesthetization in 0.15% 3-aminobenzoate, an abdominal incision was made to perform a partial oophorectomy. Oocytes were manually dissected from the ovarian lobes and incubated in Ca2+-free OR2 medium (in mM: 82.5 NaCl, 2 KCl, 1 MgCl2, and 5 HEPES) containing collagenase (type IA, 2 mg/ml; Sigma) for ∼2 h at room temperature to remove the follicular layer. After thorough washes in collagenase-free OR2 medium, oocytes were stored in a modified L15 medium (50% Leibovitz's medium, 10 mM HEPES, pH 7.4) at 19°C. Injection of healthy-looking Dumont stage V–VI oocytes was performed 12–14 h after isolation with a pneumatic PV829 picopump (World Precision Instruments). After injection of 0–50 ng of cRNA, oocytes were stored in L15 medium at 19°C and channel activity was assessed 1–3 days after injection.
Recording solutions and electrophysiology.
Whole cell oocyte currents were measured using a two-microelectrode voltage clamp equipped with a bath-clamp circuit (OC-752B; Warner, New Haven, CT) as described previously (39). Voltage-sensing and current-injecting electrodes had resistances of 0.5–1.5 MΩ when back-filled with 3 M KCl. After a stable impalement was achieved, such that both electrodes measured the same membrane potential, pulse protocols were conducted. A Macintosh Centris 650 computer equipped with an Instrutech ITC16 analog-to-digital, digital-to-analog converter and Pulse software (HEKA) was used for stimulation and data acquisition. For analysis purposes, data were filtered at 1 kHz and digitized online at 2 kHz to the hard disk using Pulse and IGOR (WaveMetrics). A 45 mM K+ bathing solution [in mM: 45 KCl, 45 N-methyl-d-glucamine chloride, 1 CaCl2, 1 MgCl2, and 5 HEPES, pH 7.4] was used for two-microelectrode voltage clamp experiments. To measure plasma membrane stability, oocytes were incubated in 5 mM brefeldin A (BFA) or vehicle (EtOH) at room temperature for the specified time.
PDZ ligand of Kir 2.3 is required for basolateral membrane expression.
We found previously (22) that the extreme COOH terminus of Kir 2.3 contains a signal that is both necessary and sufficient to direct basolateral membrane expression. The unique domain juxtaposes or is colinear with a consensus sequence (XS/T X I/V-COOH) for a type I PDZ protein interaction (36), suggesting that the channel may associate with PDZ proteins in the kidney to coordinate polarized expression. To evaluate this hypothesis, we first determined the functional consequence of specifically removing the entire PDZ ligand, the COOH-terminal four amino acids, on steady-state expression of Kir 2.3 in a polarized renal epithelial cell line. Confocal images of MDCK cells stably expressing either the wild-type or truncated (442X) Kir 2.3 channels are shown in Fig. 1. In contrast to the basolateral membrane expression of wild-type Kir 2.3 and the apical expression of the 438X mutant (22), the 442X truncated channel was distributed to a punctate, vesicle-like compartment. Consistent with partial localization in recycling endosomes, a fraction of 442X channels colocalized with transferrin in a perinuclear vesicular compartment (Fig.2) but not with PDI, an endoplasmic reticulum marker, or with a lysosomal marker (Lysotracker). Collectively, these results indicate that the PDZ binding motif of Kir 2.3 is a critical determinant for basolateral membrane expression, a function conceivably achieved through an interaction with kidney PDZ proteins.
Identification and structure of hLin-7b.
To begin to elucidate the mechanism for the PDZ-dependent expression of Kir 2.3, kidney PDZ proteins that associate with the Kir 2.3 basolateral-sorting domain were identified using a directed two-stage yeast two-hybrid interaction trap screen. From a screen of ∼14 million kidney cDNAs, 116 library clones scored positively for interaction with a lexA fusion to the COOH-terminal 29 amino acids of Kir 2.3. Positive clones were clustered into common groups, based on insert size and restriction digest patterns, and subjected to a second round of screening using a lexA fusion to a mutant Kir 2.3 COOH terminus lacking the PDZ binding motif (442X). Five cDNA clones that interacted with the wild-type bait failed to interact with the mutant, consistent with the presence of a PDZ domain (Fig.3). Sequence analysis revealed that two of the five clones were overlapping cDNAs encoding the same gene product. A BLAST search (1) of public databases revealed that the two overlapping cDNAs share a high degree of similarity to aC. elegans protein called Lin-7 (18). Because this is the second human Lin-7 homolog to be identified, we named it hLin-7b. An amino acid alignment of hLin-7b, VELI2/MALS2 (the corresponding mouse form; Refs. 4, 16), andC. elegans Lin-7 is shown in Fig.4 A. Excluding the longer and divergent NH2-terminal domain in the C. elegansprotein, hLin-7b shares 84% identity with the C. elegansLin-7. Supporting the existence of a family of related isoforms in humans, the primary structure of hLin-7b is more similar to the corresponding mouse homolog, Veli2/MALS2, than the other human Lin-7a form (97% vs. 67% amino acid identity).
All of the Lin-7 proteins identified to date are composed of two conserved domains; the COOH-terminal segment encodes a PDZ domain, whereas the NH2-terminal region contains a highly conserved domain that has been shown to mediate Lin-2/CASK interaction (19,38). COILS software analysis (23) revealed that the conserved NH2-terminal region in Lin-7b and Lin-7c has a high probability of forming a short coiled coil, consisting of two heptad repeats (Fig. 4 B). Interestingly, a single amino acid substitution, a cysteine at position 15, disrupts the coiled coil in one of the mammalian isoforms, Lin7c/Veli3/MALS3, perhaps indicating different binding preference for CASK or related Lin-7 binding proteins (19). A conserved region of low compositional complexity was identified in all forms (SMART; Ref. 32) between the PDZ and the NH2-terminal CASK interaction domain. Thought to be nonglobular in nature, the region may encode a flexible hinge structure that punctuates the two protein-protein interaction domains, presumably allowing both interaction domains to function simultaneously (Fig. 4 C).
Recognition sequence for hLin-7b PDZ domain.
The basolateral membrane sorting signal of Kir 2.3 lies in close proximity to and may even overlap with the PDZ binding motif. We recently found (22) that Kir 2.3 Δ431–441 channels, harboring an internal deletion mutation that leaves the last four amino acids intact, are mistargeted to the apical membrane, consistent with the removal of a basolateral sorting determinant. To determine whether this upstream sorting domain influences hLin-7b interaction, the target sequence for hLin-7b binding was delineated using the yeast two-hybrid system. A series of alanine substitution mutations were created in lexA-Kir 2.3 COOH terminus and tested for their ability to interact with the AD-hLin-7b protein. All baits were expressed in yeast at relatively equal levels as determined by Western blot analysis (Fig.5 B). To quantify the relative strength of interaction, reporter activity (β-galactosidase) was measured using the ONPG solution assay. Counting from the extreme COOH-terminal amino acid of Kir 2.3, residues at positions 0, −2, and −3 were found to be absolutely required for interaction with hLin-7b. Alanine substitution at these sites specifically rendered the channel unable to interact with hLin-7b (Fig. 5 A), consistent with a classic type I PDZ interaction. In contrast, point mutations made further upstream, beyond the −3 residue, had no significant effect on the interaction between lexA-Kir 2.3 and AD-hLin-7b proteins. The result indicates that individual residues in the upstream sorting determinant do not significantly influence Lin-7 interaction with Kir 2.3. In fact, replacement of the entire sorting domain (431) with an 11-residue stretch of alanines (431–41A) attenuated but did not eliminate binding of hLin-7b to Kir 2.3 (Fig. 5 A). Collectively, the observations provide reason to speculate that the Lin-7 binding site may be distinct from the biosynthetic sorting signal.
Coimmunoprecipitation of hLin-7b and Kir 2.3.
To authenticate the interaction of hLin-7b with Kir 2.3 in mammalian cells, the full-length proteins were tested for association with one another in COS cells by coimmunoprecipitation. For this purpose, Kir 2.3 and hLin-7b were tagged with the VSV G protein and HA epitopes, respectively. Immunoblotted lysates from COS cells transiently transfected with each of the tagged proteins were found to specifically express a 29-kDa protein corresponding to HA-hLin-7b or a 53-kDa protein corresponding to VSV-Kir 2.3 (Fig.6, lanes 2).
A Kir 2.3/hLin-7b complex could be immunoprecipitated from COS cells cotransfected with both proteins. Immunoprecipitation of HA-hLin-7b and associated proteins was performed using a rabbit anti-HA antibody. Recovered immunoprecipitates were immunoblotted with anti-VSV antibody, specifically recognizing coimmunoprecipitated VSV-Kir 2.3 at 53 kDa (Fig. 6 B, lane 5). As controls, coimmunoprecipitation of VSV-Kir 2.3 with the HA antibody required cotransfection of VSV-Kir 2.3 and HA-Lin-7b; no products were identified in cells that were transfected with VSV-Kir 2.3 or HA-Lin-7b alone (Fig. 6 B, lane 6). Additionally, cotransfected COS cells subject to immunoprecipitation without antibody (Fig. 6 B, lane 4) or with a control rabbit antibody (data not shown) also failed to immunoprecipitate VSV-Kir 2.3. These results confirm that the full-length hLin-7b protein is capable of interacting with full-length Kir 2.3 channel in mammalian cells.
hLin-7b promotes plasma membrane retention of Kir 2.3.
To investigate the functional consequences of hLin-7b interaction with Kir 2.3, we examined macroscopic Kir 2.3 channel activity inXenopus oocytes using the two-microelectrode voltage clamp. Coinjection of hLin-7b cRNA significantly increased the macroscopic Kir 2.3 conductance, indicating an increase in active channel number, single-channel conductance, or open probability. (Fig.7, A and B). To test whether the response involves an increase in plasmalemma stability, oocytes were exposed to BFA or vehicle (EtOH) and macroscopic currents were measured over 9 h at the time points indicated in Fig. 7 C. BFA is a fungal metabolite that inhibits transport of newly synthesized membrane proteins to the plasmalemma by blocking anterograde trafficking of vesicles from the endoplasmic reticulum to the Golgi apparatus. Consequently, the lifetime of active channels on the plasma membrane can be ascertained by monitoring channel activity in the presence of BFA. Exposure to BFA caused a progressive decrease in macroscopic channel activity, well described by a single exponential with a half-life of 4.3 ± 0.85 h. The response occurs with a time course similar to that of clathrin-dependent removal of other renal collecting duct channels, epithelial Na+ channel (ENAc; Ref. 34) and renal outer medullary K+ channel (ROMK; Ref.42), as expressed in Xenopus oocytes. Although they are suggestive, it should be pointed out these residence time measurements do not provide a strict quantification of endocytosis. It is probable that recycling of endosomes back to the plasmalemma occurs, causing an underestimate in the rate of endocytosis. Nevertheless, and in dramatic contrast to these results without hLin-7b, Kir 2.3 channel activity was not affected by BFA in oocytes coinjected with hLin-7b (Fig. 7 C). Collectively, these results are consistent with our hypothesis that hLin-7b stabilizes Kir 2.3 channels in the plasma membrane either by retention or by promoting recycling from early endosomes.
hLin-7b associates with CASK in MDCK cells.
In C. elegans, Lin-7 directly interacts with Lin-2, forming a part of a multimeric basolateral scaffold that coordinates basolateral membrane expression of the Let-23 receptor (18). To determine whether hLin-7b interacts with a related protein in mammalian epithelia, MDCK cells were stably transfected with HA-hLin-7b and examined for the presence of associated proteins by HA immunoprecipitation. CASK, a mammalian ortholog of Lin-2, copurified with HA-hLin-7b on HA antibody-agarose beads (Fig.8 A, lane 5). As controls, CASK could not be immunoprecipitated with the HA antibody from wild-type cells (Fig. 8 A,lane 4) or with agarose beads alone from HA-hLin-7b-transfected MDCK cells (Fig. 8 A, lane 3). Supporting the idea that hLin-7b associates with a specific PDZ complex, syntrophin, an unrelated basolateral membrane PDZ protein (17), did not copurify with HA-hLin-7b (Fig.8 B, lane 5). Collectively, these data convincingly establish that the Lin-7/Lin-2 (CASK) complex is conserved in mammalian epithelia.
hLin-7b/CASK complex localizes to basolateral membranes of MDCK cells.
To provide insight into the functional roles of the hLin-7b interaction with Kir 2.3, laser scanning confocal microscopy was used to examine the subcellular expression pattern of hLin-7b. As shown in Fig.8 C, the HA-hLin-7b protein localized to the basolateral membrane, similar to the distribution of Kir 2.3. The basolateral membrane expression pattern of hLin-7b was consistently observed in three different clonal lines of HA-hLin-7b-transfected MDCK cells. As predicted by the coimmunoprecipation studies, CASK also predominantly localized to the basolateral membrane in MDCK cells (Fig.8 D). The steady-state basolateral membrane expression of these proteins suggests that the Lin-7/Lin-2 complex acts to stabilize proteins like Kir 2.3 at the basolateral membrane.
Kir 2.3 and mLin-7/CASK complex localize to basolateral membrane of CCD principal cells.
To test whether Kir 2.3 might interact with the Lin 7/CASK complex in the kidney, immunolocalization studies were performed. First, the specificity of the anti-Kir 2.3 antibody was tested by Western blot analysis with COS cells that were transfected with either VSV-Kir 2.3 or a related collecting duct Kir channel, Kir 1.1, tagged with the FLAG epitope. As shown in Fig.9 A, the anti-Kir 2.3 antibody exclusively detects an appropriately sized 59-kDa protein in COS cells transiently transfected with VSV-Kir 2.3 (lane 2). In contrast, the anti-Kir 2.3 antibody did not detect the FLAG-tagged ROMK. The anti-Kir 2.3 antibody also was found to recognize a single band corresponding to the predicted size of Kir 2.3 in the membrane fraction of kidney homogenates (Fig. 9 B,lane 2). Collectively, these results establish that the anti-Kir 2.3 antibody specifically recognizes Kir 2.3.
Kidney sections colabeled with the anti-Kir 2.3 antibody and an antibody to the aquaporin 3 (AQP3) water channel, a marker for the CCD basolateral membrane, reveal predominant expression of the Kir 2.3 channel at the basolateral membrane of CCD principal cells (AQP3-positive cells; Fig. 9 C), as well as diffuse distribution throughout the cytosol. Consistent with a Kir 2.3 PDZ retention complex, CCD principal cells also express mLin-7 and CASK at the basolateral membrane (Fig. 9 D). As further evidence for a partnered relationship between these proteins, Kir 2.3, mLin-7, and CASK antibodies also label glomeruli (Fig. 9, E andD, respectively). Combined with the coimmunoprecipitation data above, the basolateral membrane expression of Kir 2.3 and the mLin-7/CASK complex in collecting duct provides strong support for physiological relevant interaction in vivo.
We found recently (22) that the extreme COOH terminus of the inwardly rectifying K+ channel Kir 2.3 is both necessary and sufficient to direct basolateral membrane expression. In the present study, we have elucidated one mechanism in a multistep sorting process. Mutagenesis studies indicate that the COOH-terminal sorting domain instructs at least two distinct functions. Kir 2.3 Δ431–441 channels, harboring an internal deletion mutation that leaves the archetypal PDZ binding site intact, are mistargeted to the apical membrane (22), consistent with the removal of a basolateral sorting determinant. In contrast, removal of the neighboring PDZ binding site produced channels that localize within an intracellular vesicle compartment. In this respect, the PDZ ligand is required for efficient plasma membrane expression, dictating delivery, retention, or both. Our discovery of an ortholog of a C. elegans PDZ protein, hLin-7b, that interacts with the COOH-terminal tail of Kir 2.3 identifies one component of the sorting machinery and provides compelling evidence for a retention mechanism in a hierarchical basolateral trafficking program.
Lin-7 was originally identified in C. elegans(35) as member of a gene set (lin-7,lin-2, and lin-10) required for vulva progenitor cell (VPC) differentiation (18). Together, the products of these three genes, all containing PDZ domains, form a heterotrimeric protein complex that coordinates the basolateral membrane expression of Let-23, a tyrosine kinase receptor (18, 37). Oligomerization of the complex is specified by unique sites within each gene product, freeing the PDZ domains and other protein-protein interaction domains to recruit membrane proteins (4) and cytoskeletal elements (6) into a multiprotein complex. With the possibility for differential assembly, the modular design of the system may provide a mechanism for diversity (see below). In VPC, Lin-7 acts as the upstream scaffolding molecule, binding directly to the Let-23 receptor through a type 1 PDZ interaction and with Lin-2 via a unique NH2-terminal interaction (18, 35). Our discovery that the Kir 2.3 channel interacts with hLin-7b in mammalian renal epithelia suggests a similar mechanism for polarized basolateral membrane expression.
Orthologous gene products (Lin-7 = Veli/MALS; Lin-2 = CASK; Lin-10 = Mint/X11) were recently identified as a native tripartite complex in the mammalian brain (4), supporting the notion of an evolutionarily conserved mechanism for compartmentalizing proteins at specific membrane domains. On the basis of known binding capacities of each component of the complex, it has been suggested that the complex might act in two different steps of a membrane protein-targeting program in neurons (3, 4). Interestingly, KIF17, a neuron-specific molecular motor in neuronal dendrites, was recently found to interact with the PDZ domain of mLin-10 (Mint1/X11), providing a mechanism for selective transport of vesicles containing the complex (33). Lin-10 also interacts with Munc-18, suggesting a link to the SNARE machinery (4). Once docked, Lin-7, CASK, and Lin-10 would then act as a retention complex, recruiting different membrane proteins into clusters at the membrane via different types of PDZ interactions. For instance, the PDZ domain in CASK binds the synaptic adhesion molecule neurexin (4), potentially anchoring the complex at the presynaptic junction.
Although this is an intriguing hypothesis for protein targeting and retention in neurons, the biology of Lin-10 and Munc-18 in the kidney provides a reason to suspect a significant divergence in mammalian epithelia. The mammalian counterpart of Lin-10 is actually encoded by a family of proteins called the Mints or X11s (28). Although all three members of the Mint family share COOH-terminal PDZ and PTB domains, only the neuron-specific form, Mint-1, contains a CASK interaction domain (3), suggesting that the Mint module is dispensable for polarized targeting of the mammalian kidney. The alternative view that an undefined Mint isoform assembles with CASK in the kidney cannot be ruled out. However, because Munc-18 has been implicated in apical rather than basolateral membrane sorting (31), such a model would also necessitate a completely different intracellular routing mechanism than has been proposed for synaptic vesicle trafficking. In the absence of the Mint-1 module, the system loses the obvious link to vesicular trafficking. Subsequently, the Lin-2/Lin-7 complex may play a restricted role in retention of proteins at the plasmalemma.
Our data support such a Lin-7/CASK anchoring complex. As shown by coprecipitation, CASK interacts with hLin-7b in MDCK cells. As predicted for a retention complex, immunolocalization studies revealed that CASK and hLin-7b are largely coexpressed on the basolateral membrane of MDCK cells and renal epithelia (38). Our observation differs from the predominant intracellular localization of CASK in some neuronal cells, where it has been implicated in vesicular trafficking (33) or nuclear translocation (15). In contrast, CASK has been shown to bind the heparin sulfate proteoglycan syndecan on the basolateral membrane of mammalian epithelia via a type II PDZ interaction and with the actin/spectrin binding protein 4.1 (6). By linking extracellular matrix receptors and the cytoskeleton, the CASK complex provides a stable anchor to retain mLin-7 interacting proteins, like Kir 2.3, on the basolateral membrane.
Recent studies on the γ-amino butyric acid transporter (BGT-1) support the view that mLin-7 regulates the surface density of particular transport proteins on the basolateral membrane of renal epithelia. Like Kir 2.3, the COOH-terminal tail of BGT-1 conforms to the optimal binding motif for mLin-7 interaction (D/E S/T X I/V-COOH) (29). Perego et al. (29) demonstrated that removing the PDZ ligand (BGTΔ5) disrupted mLin-7 association in MDCK cells and dramatically increased the internalization of the transporter from the plasmalemma. Significantly, BGTΔ5 transporters are predominantly localized on the basolateral membrane despite the increase in endocytotic retrieval. In this regard, the PDZ binding domain in BGT-1 appears to operate solely as a retention signal without effecting a basolateral trafficking step.
Although the general observations with BGT-1 are comparable with those with Kir 2.3, a distinct feature of the Kir 2.3 442X phenotype suggests disparate sorting mechanisms. In contrast to BGT-1, truncating the PDZ binding motif in the Kir 2.3 channel produced a predominant, if not exclusive, intracellular expression pattern. The observation indicates that the endocytic retrieval of the truncated channel might be more avid than the transporter; an intracellular routing step of the channel may be coordinated by a PDZ interaction or both. The two different, but not mutually exclusive, explanations should be considered.
Because PDZ interactions can control multiple intracellular routing and plasmalemma retention steps (7), the 442X phenotype simply provides insight into the nature of the compartment where the PDZ binding signal is first processed. Like BGT-1, the COOH-terminal domain of Kir 2.3 contains a di-leucine motif and a classic tyrosine-based internalization signal, supporting the potential for clathrin-mediated endocytosis from the plasmalemma as well as direct endosomal targeting from the trans-Golgi apparatus (2). Indeed, our observation that the truncated 442X channel partially colocalized with transferrin in perinuclear vesicles is consistent with trafficking into an endosomal recycling compartment (40). Unfortunately, the lack of free amino or carbohydrate groups in the small extracellular domain of Kir 2.3 precludes biotinylation targeting studies to easily determine how the 442X channel reaches the endosome. Accordingly, we cannot be certain whether the first PDZ-dependent processing step curbs endosomal trafficking from the TGN or from the plasmalemma. Moreover, we cannot exclude that possibility that docking and fusion of Kir2.3-containing endosomes occurs in a PDZ-dependent fashion. Clearly, all of these steps may normally control plasma membrane expression of the wild-type channel.
Nevertheless, our identification and characterization of Lin-7 as an interaction partner of Kir 2.3 illuminates one PDZ-dependent trafficking step. As mentioned above, steady-state expression of the Lin-7 complex on the basolateral membrane is consistent with a plasmalemma retention system. Furthermore, the effect of Lin-7 on Kir 2.3 in oocytes provides direct evidence for PDZ-dependent plasmalemma stabilization. Certainly, coexpression of Lin-7 dramatically increased the lifetime of active Kir 2.3 channels on the plasma membrane. Future studies will be required to elucidate the actual retention mechanism, particularly to determine whether Lin-7 promotes recycling from early endosomes or directly secures the channel on the basolateral membrane.
In conclusion, a functional PDZ binding motif juxtaposes or is colinear with the basolateral sorting signal in the Kir 2.3 channel. Our discovery that an ortholog of a C. elegans PDZ protein, hLin-7b, interacts with the COOH-terminal tail of Kir 2.3 identifies one component of the sorting machinery and provides compelling evidence for a retention mechanism in a multistep basolateral trafficking program.
We acknowledge the expert technical assistance of Jie Liu. We thank D. S. Bredt for the anti-Kir 2.3 antibody and B. Margolis for the anti-Lin-7 antibody.
This study was supported in part by funding from the American Heart Association (P. A. Welling, Established Investigator), National Institute of Diabetes and Digestive and Kidney Diseases Grants DK-54231 to P. A. Welling and DK-32839 to J. B. Wade, and a North American Treaty Organization travel award to J. Merot and P. A. Welling.
Address for reprint requests and other correspondence: P. A. Welling, Dept. of Physiology, Univ. of Maryland School of Medicine, 655 W. Baltimore St., Baltimore, Maryland 21201 (E-mail:).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
First published September 21, 2001; 10.1152/ajpcell.00249.2001
- Copyright © 2002 the American Physiological Society